User Tag List

Page 3 of 7 FirstFirst 1 2 3 4 5 ... LastLast
Results 21 to 30 of 62

Thread: The Clickteam Fusion 2.5 optimisation & performance Hard Data thread

  1. #21
    Clicker Fusion 2.5Fusion 2.5+ DLC
    casleziro's Avatar
    Join Date
    Mar 2013
    Location
    United States
    Posts
    643
    Mentioned
    11 Post(s)
    Tagged
    0 Thread(s)
    @Schrodinger: I get the same results with both options as well. Perhaps the performance boost is specific to certain graphics cards?

  2. #22
    Clicker Fusion 2.5 DeveloperAndroid Export ModuleHTML5 Export ModuleSWF Export Module
    happygreenfrog's Avatar
    Join Date
    May 2011
    Location
    I.L.T.D.O.I.R (I.L.T.D.O.I.R's Location: The Dimension Of Infinite Recursion)
    Posts
    4,307
    Mentioned
    4 Post(s)
    Tagged
    0 Thread(s)
    To be fair, I can't reproduce the anti-aliasing performance boost in Quadratic Attack 3D, so I think there may be another factor at play that isn't the graphics card. As for War for Robovania, the game makes heavy usage of alpha channels... maybe alpha channels are an important factor here? I don't think I used those much in Quadratic Attack 3D...

    *does some testing with schrodinger's MFA*

    Okay, alpha channels don't seem to be a factor.

    Double the object count (both in the frame settings and the object creation group), disable fine collision detection, disable menu bar:

    Without anti-aliasing checked: Stabilizes at 11FPS
    With anti-aliasing checked: Stabilizes at 11FPS

    Resize display to fill window size+anti-aliasing while resizing: No effect.

    Okay, now I'm clueless.

    Let me perform a stress-test on War for Robovania, though...

    Test without anti-aliasing: 34-38FPS. Seems about right to me.

    Same test with anti-aliasing: 25FPS? Okay, something funny was going on there, because now it's running at 40-42FPS after minimizing and re-opening the game, which is also what it was running at before a mysterious huge lag spike.

    To summarize the test results: 34-38FPS without anti-aliasing, 40-42FPS with an odd lag spike down to 25FPS with anti-aliasing.

    So, based on my not-so-in-depth tests, there seems to be something​ that you can do that will make it give you a performance boost... now we just need to nail down what​.

    BTW, the anti-aliasing setting was only enabled for player projectiles, enemy projectiles, enemies, and the visible part of player units (they have a separate hitbox object because reasons, with said object not having anti-aliasing enabled). By changing the setting for more objects, I could possibly squeeze out an even better framerate.

    And upon doing further test... it seems it can sometimes hinder the framerate. Either that, or the tests I've been doing are too unreliable to be a good indicator of performance. Considering the nature of the game, which means I'm likely doing something even slightly different which could affect performance drastically, I'm leaning towards the latter.

    Tl;dr War for Robovania results are a bit more inconclusive than I would've hoped.

  3. #23
    Clicker Fusion 2.5 DeveloperFusion 2.5+ DLCAndroid Export Module
    JoshMakesGames's Avatar
    Join Date
    Jul 2016
    Posts
    108
    Mentioned
    1 Post(s)
    Tagged
    0 Thread(s)
    Checking the anti-aliasing box can improve performance?

    Huh. I'll have to try it out sometime to see for myself.

  4. #24
    Clicker Fusion 2.5 DeveloperAndroid Export ModuleHTML5 Export ModuleiOS Export Module
    Fusion 2.5 (Steam)
    schrodinger's Avatar
    Join Date
    Nov 2014
    Posts
    3,155
    Mentioned
    27 Post(s)
    Tagged
    1 Thread(s)
    Thanks for doublechecking casleziro and happygreenfrog!

    Btw, I also had some odd drops while testing that example
    similar to what happygreenfrog noticed,
    I got 28/29 fps at first, but then they dropped to 20/21
    without me making anything else than switching the test off/on a couple times..

    If anyone could make up a "working" file, simple enough,
    with antialiasing giving that boost,
    would be awesome to make some other testing

  5. #25
    Clicker Fusion 2.5 (Steam)Fusion 2.5 Developer (Steam)Android Export Module (Steam)HTML5 Export Module (Steam)iOS Export Module (Steam)
    janpier's Avatar
    Join Date
    Dec 2014
    Posts
    96
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Hi Casleziro, I saw your example to optimize the game, could you help me with my ?? I noticed that objects with a bigger object block themselves, even on my computer, where they should run smoothly. Looking for a while running the objects you should see the defect, and if my computer locks up then I imagine on the mobile devices, however, they are still only tubes, nothing specialAttachment 22902Application ex_2.2 (1) (1).mfa

  6. #26
    Clicker Fusion 2.5
    Del_Duio's Avatar
    Join Date
    Sep 2008
    Location
    Cygnus X-I
    Posts
    919
    Mentioned
    1 Post(s)
    Tagged
    0 Thread(s)
    First off, THANK YOU Vol this is a much needed thread!!!

    One thing I noticed though, at least with my computer:

    NO IMPACT
    "using an alpha coefficient (translucency) VS not using it"

    I can tell you that with my Just a Cleric game I had an active that had different shades on an alpha channel that was put over a regular backdrop picture. I did this because it had a very cool shading effect for the underground section in the main town. This severely slowed down my game whenever it went to scroll from right to left. When I removed this alpha channel active object it ran normally again. My solution was to use the alpha channel again, take a screenshot, and then crop the screenshot for the new background picture.

    But as you can see that's a hell of a way around it and a big pain in the ass if you need to change graphics later for whatever reason.

    --

    But I've also had some mighty performance issues before with some computers over others so who knows? This might not effect most people but I thought I'd throw this info out there.

  7. #27
    Clicker Fusion 2.5 (Steam)Fusion 2.5 Developer (Steam)Fusion 2.5+ DLC (Steam)Android Export Module (Steam)HTML5 Export Module (Steam)iOS Export Module (Steam)Universal Windows Platform Export Module (Steam)
    Volnaiskra's Avatar
    Join Date
    Jan 2014
    Location
    www.sprykegame.com
    Posts
    2,238
    Mentioned
    86 Post(s)
    Tagged
    0 Thread(s)
    Quote Originally Posted by schrodinger View Post
    @ Volnaiskra: I'm failing to reproduce the antialiasing boost,
    this super-simple file gives same results both with antialiasing turned on and off:

    Attachment 22871

    (just wait few seconds for the framerate to drop, at least, in my average desktop machine goes around 20)
    (p.s. tried with moving bouncing ball objects too, same result)

    Re-reading your original post,
    is it required for the object to be moved in a fastloop for this increase to show?
    I had a play with your mfa, but it just generated more questions, and no answers. First off, do you have any graphics card in that desktop machine? Because when I ran your MFA, I didn't get any framerate drop. Instead, I got a steady 530fps [flexes bicep] (I had to increase the MFA's framerate to 1000 to see this limit). I'd hate to see what FPS you get in Spryke! But yes, I get 530 whether AA is turned on or off.

    I speculated that the objects should indeed be moving for the AA to make a difference, since that would force Fusion to have to constantly redraw each object. I also wondered whether your object, which was created within Fusion and didn't contain any translucent pixels in it, was the issue. So with that in mind, I made several adjustments to your MFA: I increased the number of objects, increased resolution, imported a PNG32 from photoshop, made all objects move in a fastloop (made them smaller so they could move more without going off screen - going offscreen improves performance unintentionally), and made the camera move independently on top of that. I got it down to where it runs at about 25fps on my maching. Still, no matter what I did, I could produce no discernible FPS difference between AA and non-AA. Peculiar!

    However, I noticed two other strange things:

    (1) Somewhere along the process of modifying your MFA, its utilisation of my GPU usage went from almost full to almost none!! In your version, once I unlocked the 1000 framerate, it pushed my GPU to 97%. My version of the MFA (attached) only uses 1% of the GPU! I don't actually know what change of mine is responsible for this, but it is very interesting. Perhaps the use of fastloops so stresses the CPU that the GPU no longer has any significant workload?

    (2) I was unable to produce any difference between AA and non-AA. However, in both your MFA and my MFA, I noticed a significant difference when the "transparent" setting was changed (also in the "effects" section). As with my original "antialiasing" findings, this one is counter intuitive because the ostensibly cheaper option is actually the one that is expensive: setting "transparent" to ON gives a performance boost. And again, there is no discernible (for me) visual impact of this setting: it changes the appearance of the active object in the editor, but not in the runtime. In my MFA (attached), turning "transparent" OFF decreased FPS from 25 to 21. In Schrodinger's MFA, it decreased it from 530 to 260!!! Interestingly, I'm pretty sure I tested this myself earlier and found no difference (I can speculate again that perhaps my extensive use of fastloops in those previous tests bottlenecked the CPU so much that this GPU-dependent performance saving was made irrelevant). Can you guys please test this yourselves and report back (using either Schrodinger's MFA quoted above or my own attached one)?




    Meanwhile, I'll set about recreating the original test I did for antialiasing (I never saved it). Hopefully it'll get similar performance savings again, in which case I'll upload it, and we can try and dissect what's going on.
    Attached files Attached files

  8. #28
    Clicker Fusion 2.5 (Steam)Fusion 2.5 Developer (Steam)Fusion 2.5+ DLC (Steam)Android Export Module (Steam)HTML5 Export Module (Steam)iOS Export Module (Steam)Universal Windows Platform Export Module (Steam)
    Volnaiskra's Avatar
    Join Date
    Jan 2014
    Location
    www.sprykegame.com
    Posts
    2,238
    Mentioned
    86 Post(s)
    Tagged
    0 Thread(s)
    Quote Originally Posted by Del_Duio View Post
    One thing I noticed though, at least with my computer:

    NO IMPACT
    "using an alpha coefficient (translucency) VS not using it"

    I can tell you that with my Just a Cleric game I had an active that had different shades on an alpha channel that was put over a regular backdrop picture. I did this because it had a very cool shading effect for the underground section in the main town. This severely slowed down my game whenever it went to scroll from right to left. When I removed this alpha channel active object it ran normally again. My solution was to use the alpha channel again, take a screenshot, and then crop the screenshot for the new background picture.
    Actually, this doesn't sound like it refutes the "using an alpha coefficient (translucency) VS not using it" result, because that was about entering a coefficient into the box shown below. In both tests, the same active objects were present, but in test A they had a coefficient of 0, and in test B it was 150 (150 makes the entire active object roughly half-transparent)



    However, this may refute another of my results: that switching between PNG8/PNG24/PNG32 has no impact, since we're talking about images that contain alpha channels (ie. PNG32) vs images that don't (eg. PNG8). If so, Danny will be delighted!

    But from what I understand of your description, it sounds like there is more than one factor that could be at play here. If I understand correctly, your scenario looked like this:

    TEST A:
    backdrops
    actives (containing images and alpha channel)

    TEST B:
    backdrops only (with an effect pre-baked into the backdrop)

    So maybe it was simply the addition of the extra active objects (resulting in more total objects) that caused the slowdown (were there many of them?) Or it could have revealed that backdrops are faster than actives. Or both. If you're easily able to test in a way that strictly isolates the alpha channel (ie. using the same number of actives in both tests, with only one set containing an alpha channel) then could you report your findings?

  9. #29
    Clicker Fusion 2.5
    Del_Duio's Avatar
    Join Date
    Sep 2008
    Location
    Cygnus X-I
    Posts
    919
    Mentioned
    1 Post(s)
    Tagged
    0 Thread(s)
    That's probably what it was, Vol.

    I do know that since that shady active has been gone I was able to add a lot more active objects without an alpha channel and there was no further slowdown like before (ex. the little NPCs under the ground behind the secret walls and stuff, the squirrel storage service up above and etc).

    If I had to guess since this build is long long gone by now I'd say the active with the different shades on the alpha channel was approx. 120px wide by 60px high. Maybe it was the size of the active that was causing the issues, since all the NPCs and stuff that came later are much smaller at 16x16px.

  10. #30
    Clicker Fusion 2.5 DeveloperAndroid Export ModuleHTML5 Export ModuleiOS Export Module
    Fusion 2.5 (Steam)
    schrodinger's Avatar
    Join Date
    Nov 2014
    Posts
    3,155
    Mentioned
    27 Post(s)
    Tagged
    1 Thread(s)
    @ Volnaiskra: tried your latest mfa,
    got a very very small gain (+1 fps) with "transparent" ticked (from 14/15 to 15/16 fps)

    So it seems we're a bit puzzled on these "ticks" performance gains.
    Your considerations are probably the answer: there's likely a lot of "balancing" work under the hood we can hardly grasp,
    even in such simple test-situations we're trying to create, the runtime, the processor(s), threads handling, our computers different hardware and services,
    will (should) all try to squeeze the best out of our apps, and this can all make for highly evasive answers to our questions.

    eheheh I see you probably have a quite good machine!
    I did test Spryke demo some time ago on this pc,
    remember experiencing/reporting some jitteriness in your WIP thread,
    don't know if you did significant optimizations since then.

    My specs below:

    PROCESSOR INFORMATION
    Manufacturer Intel
    Model Intel(R) Core(TM) i5-4250U CPU @ 1.30GHz
    CPU Speed 1.9 GHz

    GRAPHICS INFORMATION
    Intel®HDGraphics5000
    3D Acceleration Yes
    Hardware Transform & Lighting Support Yes
    Video Memory 1.7 GB
    Vertex Shader Support 5.0
    Pixel Shader Support 5.0
    Microsoft DirectX* Version 11.0

    it's an average i5 desktop with SSD drive and low-energy components (a bit like ultrabooks),
    suitable for coding, office, browsing, low-to-medium-impact graphics.. surely not for modern gaming.

    But it's probably a good idea for devs using an average machine as their "core" tool for coding,
    unless they primarily work on the graphic side and make strong use of heavy graphic packages,
    or unless they're developing a super-powered 3D game for NASA computers, of course

    A powerful machine could give bad surprise when broadly testing the app,
    while a total crap one could make most of editing job a pain!

Page 3 of 7 FirstFirst 1 2 3 4 5 ... LastLast

Similar Threads

  1. Clickteam please improve PERFORMANCE of MMF2 games ASAP
    By Blue66 in forum Multimedia Fusion 2 - Technical Support
    Replies: 137
    Last Post: 12th February 2017, 05:05 AM
  2. Replies: 5
    Last Post: 22nd March 2016, 07:35 PM
  3. Optimisation Thread - getting more FPS, every known trick
    By Sebaceous in forum Multimedia Fusion 2 - Technical Support
    Replies: 37
    Last Post: 30th September 2012, 07:35 AM
  4. Data sending optimisation
    By Felik in forum Lacewing
    Replies: 11
    Last Post: 21st March 2010, 03:59 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •